Abnormality of taste and smell in Parkinson's disease

Parkinsonism and Related Disorders 15 (2009) 232e237 www.elsevier.com/locate/parkreldis

Abnormality of taste and smell in Parkinson’s disease Mussadiq Shah, Jacquie Deeb, Marina Fernando, Alastair Noyce, Elisa Visentin, Leslie J. Findley, Christopher H. Hawkes* Smell and Taste Research Unit, Essex Neuroscience Centre, Queens Hospital, Romford, Essex, RM7 0AG, United Kingdom Received 10 February 2008; received in revised form 12 May 2008; accepted 13 May 2008

Abstract Background: Smell sense is impaired in classic Parkinson’s disease (PD). An initial study found no change in taste threshold in non-demented PD subjects and pathological studies suggest that the first relay for taste, the nucleus of the solitary tract, is spared. We wished to determine if taste is abnormal in PD and whether it is associated with smell dysfunction. Methods: Taste threshold was estimated using the Rion electrogustometer and olfaction by the University of Pennsylvania Smell Identification Test (UPSIT) in 75 non-demented PD patients and 74 controls. Results: There was a significant impairment of taste threshold and severe disorder of smell identification in the PD group. Age, duration of symptoms, disability, and smoking had no important effect on threshold measurement and there was no correlation between taste and smell dysfunction. Sensitivity analysis suggested that a provisional diagnosis of PD would be confirmed if smell or taste were abnormal; conversely, the diagnosis would merit review if both modalities were normal. Conclusions: Impaired taste appreciation was found in about 27% of patients with clinically defined PD. There were no important effects from age, disease severity or smell sense. Given the sparing of the first and second order taste neurones in PD, disorder of taste in PD most likely signifies involvement of the frontal operculum or orbitofrontal cortex, in keeping with advanced disease, although confounding by drug effects and changes in salivary constitution could not be excluded completely. Ó 2008 Elsevier Ltd. All rights reserved. Keywords: Taste; Olfaction; Parkinson’s disease

1. Introduction It is well recognised that patients with Parkinson’s disease (PD) have profound disorder of olfactory function affecting at least 80% of those defined according to clinical criteria [1e3]. There is uncertainly with regard to taste. In the first large study [4], taste was assessed in 33 controls and 30 PD patients by (a) individual paper disks soaked in sucrose, quinine, citric acid or sodium chloride, (b) 100 ml samples of chocolate and vanilla milk and (c) taste threshold with the Rion TR-06 electrogustometer. They found no abnormalities in responses to pleasant or unpleasant taste stimuli or threshold in the PD group and in

* Corresponding author. Tel.: þ44 1708 435000. E-mail address: [email protected] (C.H. Hawkes). 1353-8020/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.parkreldis.2008.05.008

some, there was lowering of taste threshold on electrogustometry. A further study examined taste in 52 patients with various forms of dementia [5] of whom 6 had non-Alzheimer type dementia with parkinsonism. Taste was assessed by whole mouth and taste strip tests and the patients’ results compared to a control group consisting of 52 patients with minor strokes or vascular risk factors without stroke, some of whom had mild cognitive impairment as measured by the Minimental State Examination (MMSE). In the 6 demented PD patients, taste was impaired compared to the dementia patients without parkinsonism. This confused picture prompted further investigation. We used a simple regional test of taste (electrogustometry, EGM) in which a weak DC anodic stimulus is applied to the desired area of the tongue. Our main objective was to determine whether taste is abnormal in PD and if found, whether

M. Shah et al. / Parkinsonism and Related Disorders 15 (2009) 232e237

there is any correlation between taste and smell sense or clinical variables such as age, gender, and disease severity. 2. Methods 2.1. Control and patient selection criteria Local research and ethics committee approval and informed consent were obtained from all the participants. Controls were derived from members of hospital staff and their relatives. Some were receiving medication for hypertension or raised cholesterol level as detailed below. All participants completed a health questionnaire directed toward conditions that might interfere with smell or taste. Major reasons for exclusion were previous nasal disorder; coexistence of disease (e.g. diabetes) or drug usage. Smokers were not debarred but we did exclude those with poor oral hygiene, history of middle ear infection or Bell’s palsy. Patients receiving levodopa and other medication for PD were included. Everyone completed the MMSE, and those scoring less than 27/30 were excluded. All PD patients were in a stable motor state, fulfilled the revised criteria of the UKPD Research Group [6] and were in Hoehn and Yahr stages IeIII, i.e. less disabled.

2.2. University of Pennsylvania Smell Identification Test (UPSIT) [7] In this procedure, 40 different odours are used and a forced choice is made from 4 possible answers. The procedure was supervised in all instances; none were administered by post. No patient was observed to have difficulty in sniffing, a factor that may lower UPSIT scores slightly [8]. This method was chosen because it is simple to use, we have a large in-house healthy control dataset and it is recognised that smell identification correlates well with the other principal olfactory components of threshold and discrimination [9].

2.3. Electrogustometry The electrogustometer (Rion TR-06, Tokyo, Japan) delivers the signal by means of a sterilised stainless steel, circular electrode, 5 mm diameter applied to the upper tongue surface. We chose this approach rather than whole mouth methods as it allows regional assessment of taste appreciation and it is less tedious for patients than chemogustatory methods [10]. The signal can be varied from 6 dB to 34 dB and it is perceived variously as metallic, sour, salty or bitter. The stimulus duration was 1.5 s in all cases. Electrodes were applied separately to right and left sides of the dorsum of the tongue as follows: (a) either side of the tip, the area served by the chorda tympani (fungiform papillae, FP) (b) over the most lateral circumvallate papilla (vallate papillae, VP) on either side, the area served by the glossopharyngeal nerve. The stimulus current was increased using a single staircase approach until the subject recognised a taste sensation; any non-taste sensation reported at lower concentration was not recorded. This gives a taste recognition threshold measurement, expressed in decibels (dB) and the average of both sides was taken for each subject.

3. Results In the control group, 53 (72%) were on no medication, 21 (28%) were taking one or more preparations averaging 1.9 per subject. This comprised anti-hypertensives (8 subjects), statins (6 subjects), and antidepressants (3 subjects). Thirteen were on a variety of drugs such as asthma inhalers, thyroxine, glaucoma drops, aspirin or the oral contraceptive. In the patient group, 17 (23%) were receiving no medication, 23 (30.7%) were taking levodopa-containing medication; 4 (5%) were receiving dopamine agonists and 3 (4%) were on other PD medication (Amantidine, Selegiline, and Benzhexol).

233

The mean number of drugs taken by patients receiving PD-related treatment was 2.2. Other medication included anti-hypertensives (28 patients), statins (6 patients), and antidepressants (9 patients). The mean UPSIT score in controls was 33 (range 13e40). The mean score in patients was 19.5 (range 4e35) which is significantly lower than controls ( p < 0.0001, t-test). The mean FP threshold for controls was 11.1 dB (range 6 to 26) and for patients 19.5 dB (range 1 to 34). For VP the mean threshold for controls was 13.4 dB (range 0e31) and for patients 19.1 (range 5.0 to 34). In the control group, 26 (36%) were previous smokers; 35 (48%) never smoked and 12 (16%) were current smokers (one patient with missing data). In the PD group, 36 (49%) were previous smokers; 34 (46%) never smoked and only 4 (5%) were current smokers (1 patient with missing data). Full details of the statistical approach are provided in the Appendix. 3.1. Chorda tympani (FP) For age and gender adjusted estimates, we used tobit regression [11] of right and left averaged FP, with 5 values > 34 dB censored. Smoking adjusted estimates were not materially different from those without smoking terms in the model (smoking terms p ¼ 0.831). Patients had mean right and left (R/L) averaged FP values greater than controls by 6.9 dB (95% CI: 3.9, 9.9; p < 0.001), after adjusting for age and gender with no significant difference when adjusted for smoking. Males had mean R/L averaged FP values greater than females by 2.6 dB (95% CI: 0.05, 5.20; p ¼ 0.046) after adjusting for subject status and age, again with no meaningful difference when adjusted for smoking. There was no association between age and averaged FP in the whole subject group with an estimated increase of 0.05 dB per year ( p ¼ 0.236), however, there was borderline evidence ( p ¼ 0.082) of an interaction between age and subject status. There was also marginally significant evidence that averaged FP did increase with age in patients (by 0.21 dB per year (95% CI 0.01, 0.4; p ¼ 0.039)), but not in controls (by 0.02 dB per year (95%CI: 0.07, 0.10; p ¼ 0.706). There was no association between mean FP threshold and Hoehn and Yahr score or symptom duration, age, gender and smoking ( p > 0.05). The mean FP threshold for the 23 patients who were taking levodopa-containing drugs was not significantly different from those not taking levodopa ( p ¼ 0.7, t-test). 3.2. Vallate papillae For age and gender adjusted estimates, we used tobit regression of right and left averaged vallate papillae (VP), with 7 values > 34 dB censored. Smoking adjusted estimates were not materially different from those without smoking terms in the model (smoking terms p ¼ 0.654). Patients had a mean right and left averaged VP values greater than controls by 5.8 dB (95% CI: 2.6, 9.1; p < 0.001) after adjusting for age and gender, with no meaningful difference when adjusted for

234

M. Shah et al. / Parkinsonism and Related Disorders 15 (2009) 232e237

smoking. Males had a mean right and left averaged VP values of borderline significance that was greater than females by 2.5 dB (95% CI: 0.23, 5.31; p ¼ 0.072) after allowing for subject status and age again with no material difference when adjusted for smoking. There was no association between age and averaged VP in the whole subject group with an estimated decrease of 0.01 dB per year ( p ¼ 0.839). In this case there was no sign of a different association with age between patients and controls ( p ¼ 0.839 for the interaction test). There was no association between mean VP threshold and Hoehn and Yahr score or symptom duration when adjusted for age, gender and smoking ( p > 0.05). The mean VP threshold for the 23 patients taking levodopa containing drugs was not significantly different from the VP group mean threshold (p ¼ <0.46, t-test). 3.3. Fungiform and vallate papillae combined Age and gender adjusted estimates were analysed by tobit regression of the averaged FP and VP values, with 4 measurements > 34 dB censored. Smoking adjusted estimates were not materially different from those without smoking terms in the model (smoking terms p ¼ 0.593). Patients had mean right and left averaged combined values greater than controls by 6.1 dB (95% CI 3.3, 8.8; p < 0.001) after adjusting for age and gender, with no material difference when adjusted for smoking. Males had mean right and left averaged combined values greater than females by 2.4 dB (95% CI: 0.1, 4.9; p ¼ 0.040) after adjusting for subject status and age again with no significant difference when adjusted for smoking. There was no association between age and averaged combined FP/VP in the whole subject group (estimated increase of 0.02 dB per year, p ¼ 0.562). There was no evidence of an interaction between age and subject type ( p ¼ 0.333). 3.4. Association between taste and smell 3.4.1. Fungiform papillae Smoking adjusted estimates were no different from those without smoking terms in the model (smoking terms, p ¼ 0.858) and there was no association between mean FP and UPSIT, after adjusting for subject status, i.e. estimated mean FP in subjects with UPSIT scored higher by 1 unit fell by 0.08 dB (95% CI: 0.3, 0.1; p ¼ 0.445), after allowing for subject status, age, gender. Using conventional regression and the value 34.1 dB for censored FP values, the partial correlation coefficient between mean FP and UPSIT, after adjustment for subject status, age and gender, was 0.056 ( p ¼ 0.505). There was no difference (interaction) in FP vs. UPSIT association between patients and controls, or association between mean FP and UPSIT in patients only. 3.4.2. Vallate papillae Smoking adjusted estimates were no different from those without smoking terms in the model (smoking terms, p ¼ 0.655). There was no association between mean VP and UPSIT, after adjusting for subject status, i.e. estimated mean VP in

subjects with UPSIT score higher by 1 unit fell by 0.01 dB (95%CI: 0.2, 0.2; p ¼ 0.947), after adjusting for subject status, age, and gender. Using conventional regression and the value 34.1 dB for censored VP values, the partial correlation coefficient between mean VP and UPSIT, after adjusting for subject status, age and gender, was 0.0001 ( p ¼ 0.999). There was no difference (interaction) in VP vs. UPSIT association between patients and controls or between mean VP and UPSIT in patients only. 3.4.3. Sensitivity and specificity analysis There were 49/75 patients and 3/74 controls with an abnormal UPSIT score; and 49/52 abnormal UPSIT subjects who were patients and 71/97 with a normal UPSIT who were controls. This gives a sensitivity of 65%; specificity of 85.9%; positive predictive value (PPV) of 94.2% and accuracy of 80.5%. For taste there were 20/75 patients and 1/74 controls with abnormal FP thresholds. There were 20/21 FP abnormal subjects who were patients, 73/128 normal subjects who were controls, yielding sensitivity of 26.7%; specificity of 98.7%; PPV of 95.2% and accuracy of 62.4%. Where abnormality was defined as having both smell and taste abnormal, the sensitivity dropped markedly (18.7%); but if abnormality was defined as having either smell or taste abnormal (or both), the sensitivity increased to 73.3%, with a specificity of 94.6%; PPV of 93.2%; accuracy of 83.9%; and in this case the negative predictive value, the proportion of subjects normal on both modalities who were controls, was 77.8%. 4. Discussion The statistically significant findings of this study are as follows: (i) UPSIT score in PD patients was markedly lower than controls (ii) For FP and VP, the mean taste threshold was higher in patients than controls. Age, smoking, disability, disease duration, had no significant influence on the data. Males displayed higher thresholds than females, although both sexes were different from controls. (iii) Combining the FP and VP thresholds likewise showed that patients were inferior to controls and that male patients fared worse. (iv) There was no association between smell and taste data. (v) sensitivity and specificity analysis showed that USPIT testing had modest sensitivity (65%) but high PPV (94.5%) for a diagnosis of PD. For taste (FP) sensitivity was low (26.7%) but again PPV was high (95.2%) and remained high where abnormality was defined as having either abnormal taste or smell scores. Conversely, the probability of someone having PD where both smell and taste tests were normal was 22.2%. Our large study of taste in PD without dementia shows clear abnormalities. The defect is far from universal, affecting just 20/75 (26.7%) patients (see Fig. 1) and because of the overlap between patients and controls it could not be used alone as a diagnostic tool. The sensitivity of UPSIT is greater than taste measurement by the current technique but the combination of abnormality on both smell and taste would support a diagnosis of PD. We have no information as yet on disease specificity of the taste findings and clearly it would be of

M. Shah et al. / Parkinsonism and Related Disorders 15 (2009) 232e237

control

235

patient

Mean FP threshold (dB)

30

20

10

0

-10 20

40

60

80

100

20

40

60

80

100

Age (years) Fig. 1. To display the relation between mean fungiform papillae (FP) threshold and age in controls and patients. The regression lines are shown with 95% prediction intervals for the control group. Values above 34 dB threshold are assigned an arbitrary value of 34.1 dB.

interest to perform a similar investigation in those with ‘Parkinson’s plus’ syndromes. Many patients and a few controls were receiving medication for PD and other disorder such as hypertension, or raised cholesterol level which might elevate taste thresholds but it was difficult to control for this variable. Levodopa preparations probably have no significant effect on smell identification [12]. It is possible that levodopa containing medication could interfere with taste threshold measurement given that around 5% patients report change in taste when receiving this drug, probably due to excretion in saliva [13]. However, separate assessment of our patients who were receiving levodopa preparations, compared to those who were not, showed no difference in mean taste thresholds. Electrogustometry (EGM) is a rapid means of taste assessment and is sensitive to slight taste impairment about which the patient may be unaware. Earlier studies of EGM in 32 subjects, questioned the value of this technique, because of its poor correlation with whole mouth chemical thresholds [14]. Against this, is the well-recognised high correlation between chorda tympani lesions and raised EGM threshold [15]. There is a good correspondence between EGM threshold and the number of fungiform papillae [16] favouring the notion that lingual sensations produced by such low level currents are mediated by the taste system. More recently, a study of 114 healthy subjects revealed that EGM correlated with all taste qualities [17] supporting our use of EGM. A theoretical disadvantage of electrogustometry is that the patient may inadvertently report common sensation via the trigeminal nerve. However, the sensory threshold for lingual trigeminal receptors is higher than that for taste [16,18] and the trigeminal nerve is spared pathologically in PD [19] making it less likely that our observations were confounded by inadvertent trigeminal stimulation. The 5 mm circular electrode is a widely used standard and provides a well demarcated stimulus [18]. The stimulus duration of 1.5 s was based on a pilot study as an

appropriate duration for subjects of all ages. Shorter durations (0.5 s or 1.0 s) were insufficient to produce taste sensation at maximum current settings in some healthy elderly controls, whereas 1.5-s duration was detected by all controls, irrespective of age. Many individuals, particularly patients, found testing of the vallate papillae (VP) uncomfortable and gagging was frequent. This problem will significantly limit the general applicability of VP testing. We chose to test VP because of its supply by a different nerve (glossopharyngeal) and the possibility that more severe taste disturbance might be found in this region of the tongue. This was not the case, and in any future study it would probably be unnecessary to evaluate this somewhat inaccessible region. The lack of significant association of taste impairment, in either controls or patients, with age, smoking or olfaction is of interest. Resistance of taste to aging effects has been documented to some tastants [20]. Others have not confirmed this using threshold and supra-threshold whole mouth techniques [21]. EGM studies suggest that there is a deterioration of taste threshold from the age of 60 years for FP, and at 70 years for VP [22]. Our data imply that in contrast to olfaction, which deteriorates markedly with age, for taste there was only marginal evidence of age-related decline and only for patients not controls. This observation may relate to lack of statistical power in our study, and the consensus view is that taste appreciation does deteriorate with age and corresponds to atrophy of taste buds [21,23]. The lack of effect from smoking was unexpected. Smokers often have poor oral hygiene which may disturb taste by allowing bacterial or fungal overgrowth [24]. Only 4/75 PD patients were current smokers and just 12/74 in the control group, so our study would be underpowered. The independent abnormalities of both smell and taste in PD most plausibly relate to their separate anatomical pathways. Although patients confuse taste and smell frequently, the first and second order neurones for taste, which reside in the brainstem and thalamus, are completely separate from the olfactory

236

M. Shah et al. / Parkinsonism and Related Disorders 15 (2009) 232e237

route, which enters the medial temporal lobe and only when taste fibres reach the frontal operculum does it converge with olfactory fibres en route to the orbitofrontal cortex (OFC). In PD the solitary tract nucleus is spared [19] but the rostral part of the dorsal vagal motor nuclear complex (DMNC) which is damaged early [19], merges with the salivatory nuclei which receive afferent fibres from all three taste nerves. Thus Lewy pathology in DMNC might change the consistency of saliva with resultant deterioration in taste appreciation [25,26]. Although many patients suffer from drooling, this probably relates to defective swallowing rather than overproduction [27]. Most likely, taste impairment in PD is an indicator of frontal opercular or OFC pathology, corresponding to Braak stages 5 and 6 and advanced disability. Hence taste abnormality in PD would suggest pathologically advanced disease. In their PD group, the Polish researchers [4] describe higher intensity ratings to 0.025% quinine and lower EGM thresholds in comparison to controls. They debate whether this might relate to unusually low quinine intensity scores or high EGM threshold in their control group, or enhanced taste function by way of compensation for the olfactory defect. No firm conclusion is derived and there are insufficient raw data in their article to permit further comment. The concept of super-tasters is biologically plausible and has been described in the context of 6-n-propylthiouracil thresholds [28,29] in otherwise healthy individuals. Conversely, reduction of phenylthiocarbamide sensitivity was documented in a study of 32 male PD patients [30]. We found no evidence of EGM super-tasters as shown clearly in Fig. 1. Some have suggested that patients with isolated disorder of smell have impaired taste as measured by chemogustatory methods [31,32]. The latter authors may not have excluded completely the possibility of unrecognised hypogeusia which may be present if the cause of anosmia is peripheral, i.e. sinus disease or polyps, where there is local infection and entry into the mouth of toxic by-products of infection. In clinical practice, pure hyposmia is regularly associated with subjective complaint of impaired taste but rarely confirmed by actual measurement. We are not aware of any study that has assessed pure hyposmia and taste by EGM, which is a simpler and probably a more objective measure. Furthermore, we found that taste and smell abnormalities were independent of each other. If hyposmia caused the taste changes then the two modalities should be correlated. Nevertheless, it is conceivable that some form of central reorganisation in PD diminishes taste appreciation in the presence of smell defect. Abnormality of taste was demonstrated previously in 6 patients with PD and dementia [5] but there would be inevitable confounding by cognitive impairment. A larger study of 30 PD patients from Poland [4] found no abnormality and taste was measured with the same electrogustometer and identical stimulus parameters to our study. One reason for the difference between our respective findings is that a lower score on MMSE was accepted in their patients (24) compared with our lower limit of 27. Furthermore, control subjects were reported to be taking an average of 1.8 drugs some of which

included those for Parkinson’s disease. These confounders may have produced a higher threshold in the control group and blurred the true differences between cases and controls. In conclusion, a simple test of taste threshold on the anterior tongue may have value in confirming a diagnosis of PD in conjunction with smell identification tests. The presence of taste impairment in PD probably indicates advanced cortical disease in keeping with Braak stage 5 or 6. A normal smell test in suspected PD would raise diagnostic doubt [33] and a normal taste threshold test in the same patient would highlight the need for diagnostic review. Acknowledgement The authors thank to Dr D Altmann, Institute of Hygiene and Tropical Medicine, London for statistical support.

Appendix. Statistical Methods Handling of censored values Very high threshold values, i.e. >34 dB were off the scale of the electrogustometer. To allow this for FP and VP separately, if one of the righteleft (R/L) pairs was censored as >34 dB, the average of R/L pairs was censored if the other value was also >34 dB, or exactly 34 dB; but if the lower value was <34 dB, the censored value was treated as 34 dB to obtain the average, e.g. average of above 34 dB and 32 dB was taken as the average of 34 dB and 32 dB, whereas the average of >34 dB and 34 dB was taken as >34 dB. Similarly, for combined FP and VP averages, a censored > 34 dB value for one of the right and left averaged FP or VP was taken as 34 dB if the other average was <34 dB. Otherwise, if the second R/L pair average was 34 dB or >34, the combined average was taken also censored as above 34 dB. Statistical comparison of right and left averaged FP and VP values, and of the combined average FP and VP values, used tobit regression [11] which essentially allows for censored values >34 dB threshold taking higher values, corresponding to the tail of an appropriate Normal distribution. The assumption is that values censored by the limitations of detection follow a Normal distribution which can be estimated from the observable values. As with ordinary linear regression, age and gender covariates along with a patient indicator allows the patient vs. control difference to be estimated after adjustment for age and sex differences in patient and control groups. The tobit regression, assumes that values above the 34 dB threshold are in the tail of a Normal distribution, estimated from the values present. This has advantages over assigning an arbitrary value (e.g. 34.1 dB) to censored values, since the latter tends to underestimate the total variability in the sample (and may underestimate the p-Value by underestimating sample noise). Similar regression with UPSIT score as covariate examined the relationship between the taste and smell measures. The tobit regression assumes normality, but there was no evidence of non-normality in FP or VP values.

M. Shah et al. / Parkinsonism and Related Disorders 15 (2009) 232e237

For comparison purposes an alternative analysis used values of 34.1 dB in place of censored values, and the results of this analysis were quite similar to the tobit results. References [1] Doty RL, Deems DA, Stellar S. Olfactory dysfunction in parkinsonism: a general deficit unrelated to neurologic signs, disease stage, or disease duration. Neurology 1988;38(8):1237e44. [2] Hawkes CH, Shephard BC, Daniel SE. Olfactory dysfunction in Parkinson’s disease. J Neurol Neurosurg Psychiatry 1997;62(5):436e46. [3] Ansari KA, Johnson A. Olfactory function in patients with Parkinson’s disease. J Chronic Dis 1975;28(9):493e7. [4] Sienkiewicz-Jarosz H, Scinska A, Kuran W, Ryglewicz D, Rogowski A, Wrobel E, et al. Taste responses in patients with Parkinson’s disease. J Neurol Neurosurg Psychiatry 2005;76(1):40e6. [5] Lang CJ, Leuschner T, Ulrich K, Stossel C, Heckmann JG, Hummel T. Taste in dementing diseases and parkinsonism. J Neurol Sci 25e102006;248(1e2):177e84. [6] Hughes AJ, Ben-Shlomo Y, Daniel SE, Lees AJ. What features improve the accuracy of clinical diagnosis in Parkinson’s disease: a clinicopathologic study. 1992. Neurology 2001;57(10 Suppl. 3):S34e8. [7] Doty RL, Shaman P, Dann M. Development of the University of Pennsylvania Smell Identification Test: a standardized microencapsulated test of olfactory function. Physiol Behav 1984;32(3):489e502. [8] Sobel N, Thomason ME, Stappen I, Tanner CM, Tetrud JW, Bower JM, et al. An impairment in sniffing contributes to the olfactory impairment in Parkinson’s disease. Proc Natl Acad Sci U S A 27e3-2001;98(7): 4154e9. [9] Doty RL, Smith R, McKeown DA, Raj J. Tests of human olfactory function: principal components analysis suggests that most measure a common source of variance. Percept Psychophys 1994;56(6):701e7. [10] Ovesen L, Sorensen M, Hannibal J, Allingstrup L. Electrical taste detection thresholds and chemical smell detection thresholds in patients with cancer. Cancer 15e11-1991;68(10):2260e5. [11] Tobin JD. Estimation of relationships for limited dependent variables. Econometrica 1958;26:24e6. [12] Barz S, Hummel T, Pauli E, Majer M, Lang CJ, Kobal G. Chemosensory event-related potentials in response to trigeminal and olfactory stimulation in idiopathic Parkinson’s disease. Neurology 1997;49(5):1424e31. [13] Siegfried J, Zumstein H. Changes in taste under L-DOPA therapy. Z Neurol 1971;200(4):345e8. [14] Murphy C, Quinonez C, Nordin S. Reliability and validity of electrogustometry and its application to young and elderly persons. Chem Senses 1995;20(5):499e503. [15] Groves J, Gibson WP. Significance of taste and electrogustometry in assessing the prognosis of Bell’s (idiopathic) facial palsy. J Laryngol Otol 1974;88(9):855e61.

237

[16] Miller SL, Mirza N, Doty RL. Electrogustometric thresholds: relationship to anterior tongue locus, area of stimulation, and number of fungiform papillae. Physiol Behav 15e4-2002;75(5):753e7. [17] Ellegard EK, Goldsmith D, Hay KD, Morton RP. Studies on the relationship between electrogustometry and sour taste perception. Auris Nasus Larynx 2007;34(4):477e80. [18] Stillman JA, Morton RP, Hay KD, Ahmad Z, Goldsmith D. Electrogustometry: strengths, weaknesses, and clinical evidence of stimulus boundaries. Clin Otolaryngol Allied Sci 2003;28(5):406e10. [19] Braak H, Del TK, Rub U, de Vos RA, Jansen Steur EN, Braak E. Staging of brain pathology related to sporadic Parkinson’s disease. Neurobiol Aging 2003;24(2):197e211. [20] Murphy C. Nutrition and chemosensory perception in the elderly. Crit Rev Food Sci Nutr 1993;33(1):3e15. [21] Schiffman SS. Taste and smell losses in normal aging and disease. JAMA 22e10-1997;278(16):1357e62. [22] Nakazato M, Endo S, Yoshimura I, Tomita H. Influence of aging on electrogustometry thresholds. Acta Otolaryngol Suppl 2002; 546:16e26. [23] Mavi A, Ceyhan O. Bitter taste thresholds, numbers and diameters of circumvallate papillae and their relation with age in a Turkish population. Gerodontology 1999;16(2):119e22. [24] Ohno T, Uematsu H, Nozaki S, Sugimoto K. Improvement of taste sensitivity of the nursed elderly by oral care. J Med Dent Sci 2003; 50(1):101e7. [25] Bradley RM, Beidler LM. Saliva: its role in taste function. 31. In: Doty RL, editor. Handbook of olfaction and gustation. 2nd ed. New York: Marcel Dekker; 2003. p. 639e50. [26] Spielman AI. Interaction of saliva and taste. J Dent Res 1990;69(3):838e 43. [27] Chou KL, Evatt M, Hinson V, Kompoliti K. Sialorrhea in Parkinson’s disease: a review. Mov Disord 2007;22(16):2306e13. [28] Goldstein GL, Daun H, Tepper BJ. Influence of PROP taster status and maternal variables on energy intake and body weight of pre-adolescents. Physiol Behav 23e4-2007;90(5):809e17. [29] Prescott J, Ripandelli N, Wakeling I. Binary taste mixture interactions in prop non-tasters, medium-tasters and super-tasters. Chem Senses 2001; 26(8):993e1003. [30] Moberg PJ, Balderston CC, Rick JH, Roalf DR, Weintraub D, KleinerFisman G, et al. Phenylthiocarbamide (PTC) perception in Parkinson disease. Cogn Behav Neurol 2007;20(3):145e8. [31] Dzaman K, Jadczak M, Rapiejko P, Syrylo A, Jurkiewicz D. Assessment of the correlation between taste and smell functioning. Pol Merkur Lekarski 2005;19(111):280e2. [32] Gudziol H, Rahneberg K, Burkert S. Anosmics are more poorly able to taste than normal persons. Laryngorhinootologie 2007; 86(9):640e3. [33] Hawkes C. Olfaction in neurodegenerative disorder. Mov Disord 2003; 18(4):364e72.